How do you design a resilient message-driven architecture with Kafka?

Kafka-based resilience is achieved through partitioning, retries, idempotent consumers, and DLQs.

Producers write events to partitioned topics for scalability. Consumers process messages with offset management. Retry topics and dead-letter queues handle failures. Exactly-once semantics are achieved using idempotent producers and transactional writes.